How Brain-Inspired Computers Learn and Unlearn Information
This patent describes a method for a brain-like computer network to learn associations between different inputs and then specifically unlearn mistakes or 'false positives' using a unique feedback mechanism.
Original patent title: “Unsupervised, supervised and reinforced learning via spiking computation”
This patent describes a method for a brain-like computer network to learn associations between different inputs and then specifically unlearn mistakes or 'false positives' using a unique feedback mechanism. Granted to International Business Machines in 2022 with 23 claims, and it is expected to expire in 2039.
Coverage
What does this patent actually cover?
This patent describes a way for a special type of computer network, called a neural network, to learn by associating features from two different inputs. For example, if you show it a picture of a cat (first input) and the sound of a meow (second input), it learns to connect them. The process involves feeding the first input to a 'first neuron population' and the second input to a 'third neuron population.' As these inputs travel through the network along 'pathways,' the network extracts their features. The clever part is how it corrects errors: it 'unlearns false positives' by taking copies of 'firing events' (like signals from brain cells) from certain parts of the network and propagating them back through 'additional pathways.' This unlearning uses a 'second learning rule,' which can be an 'anti-STDP' or 'anti-Hebb' rule, effectively weakening incorrect connections.
The gap
What does this patent NOT cover?
- Does not cover traditional artificial neural networks that do not use 'spiking computation' or 'firing events' as described.
- Does not cover learning systems that correct errors without explicitly propagating 'copies of firing events' through 'additional pathways' for unlearning.
- Does not cover methods of unlearning that do not employ an 'anti-learning rule' such as anti-STDP or anti-Hebb.
- Does not cover neural networks that only process a single input without associating features between two distinct inputs.
- Does not cover systems where the unlearning process is not initiated by firing events from the second and fourth neuron populations.
These exclusions are unique to PatentBrief — derived from the actual claim language, not patent-office boilerplate.
Key facts
What made this novel
The truly novel aspect is the explicit, structured mechanism for 'unlearning false positives.' Instead of just adjusting weights during forward passes, the patent describes copying 'firing events' and sending them back through dedicated 'additional pathways' with an 'anti-learning rule' (like anti-STDP) to specifically weaken incorrect associations. This provides a targeted way to forget wrong connections.
The Patent Drawing
Schematic visualization of the patent's claim structure. Hand-drawn diagrams in progress for each landmark patent.
Where you've seen this
Real-world examples
IBM's TrueNorth neuromorphic chip
Research in low-power AI for edge devices
Brain-inspired computing hardware
Robotics learning systems requiring error correction
Why it matters
The bigger picture
This technology is important because it addresses a fundamental challenge in artificial intelligence: making learning systems more efficient and robust. By providing a specific mechanism for 'unlearning,' it allows neural networks to refine their knowledge and correct mistakes, which is crucial for complex tasks. This approach, using brain-inspired 'spiking computation,' could lead to more energy-efficient AI hardware, especially for devices that need to learn on the go.
Filed
August 16, 2019
Granted
October 25, 2022
Market context
Who's building on this
Companies in this space
International Business Machines Corp. (IBM), the assigneeassigneeThe entity that owns the patent — usually the inventor's employer or a company.Read more →, has been a leader in neuromorphic computing with its TrueNorth chip, which implements spiking neural networks. Other major technology companies and academic research institutions are also actively exploring spiking neural networks for their potential in energy-efficient AI, particularly for edge computing applications where power consumption is critical.
Market impact
This patent contributes to the growing field of neuromorphic computing, which aims to create AI hardware and software that mimic the brain's efficiency. While still an emerging area compared to conventional deep learning, advancements in unlearning mechanisms like this could accelerate the development of more robust and adaptive AI. This could lead to a new generation of AI chips and algorithms, particularly impacting sectors requiring low-power, on-device intelligence.
Claim 1 — Plain English
What this patent covers
This patent describes a way for a special type of computer network, called a neural network, to learn by associating features from two different inputs. For example, if you show it a picture of a cat (first input) and the sound of a meow (second input), it learns to connect them. The process involves feeding the first input to a 'first neuron population' and the second input to a 'third neuron population.' As these inputs travel through the network along 'pathways,' the network extracts their features. The clever part is how it corrects errors: it 'unlearns false positives' by taking copies of 'firing events' (like signals from brain cells) from certain parts of the network and propagating them back through 'additional pathways.' This unlearning uses a 'second learning rule,' which can be an 'anti-STDP' or 'anti-Hebb' rule, effectively weakening incorrect connections.
The clever bit
The truly novel aspect is the explicit, structured mechanism for 'unlearning false positives.' Instead of just adjusting weights during forward passes, the patent describes copying 'firing events' and sending them back through dedicated 'additional pathways' with an 'anti-learning rule' (like anti-STDP) to specifically weaken incorrect associations. This provides a targeted way to forget wrong connections.
What it does not cover
- Does not cover traditional artificial neural networks that do not use 'spiking computation' or 'firing events' as described.
- Does not cover learning systems that correct errors without explicitly propagating 'copies of firing events' through 'additional pathways' for unlearning.
- Does not cover methods of unlearning that do not employ an 'anti-learning rule' such as anti-STDP or anti-Hebb.
- Does not cover neural networks that only process a single input without associating features between two distinct inputs.
- Does not cover systems where the unlearning process is not initiated by firing events from the second and fourth neuron populations.
Patent timeline
Application submitted to the patent office
Application published, typically 18 months after filing
Patent officially issued
Patent enters public domain
PatentBrief Score
Impact Score
Early stage
Citation count
0/40
No citations yet
Claim breadth
15/20
Broad claimsclaimsThe numbered statements at the end of a patent that legally define what the inventor owns.Read more →
Recency
20/20
Granted within 5 years
Assignee scale
0/20
Independent or smaller assigneeassigneeThe entity that owns the patent — usually the inventor's employer or a company.Read more →
PatentBrief Impact Score — based on citation count, claim breadth, recency, and assignee scale. Not a legal assessment.
Heuristic Value Estimate
What this patent might be worth
$47K – $150K
Midpoint $94K · 12.8 yr remaining · industry ×1.6
Heuristic only — blends forward/backward citation counts, claim scope, time remaining, litigation history, and CPC-derived industry baseline. Real valuations need a professional appraisal.
Claim text not yet imported for this patent
The original legal language
Original claims
23 claims as filed with the patent office.
Concepts involved
Citations
Patent lineage
Cite this patent
Modha, D. S. (2022). How Brain-Inspired Computers Learn and Unlearn Information (U.S. Patent No. 11,481,621). U.S. Patent and Trademark Office. https://patentbrief.org/patent/us/11481621/unsupervised-supervised-and-reinforced-learning-via-spiking-computation
Auto-generated from the patent record. Double-check author order and the issue date against the official USPTO document before submitting.
Embed
Add this patent to your site
Drop this plain-English patent card into any blog post or article — free, no signup. It always links back to the full breakdown here.
<div data-patentlens-widget data-patent-number="US11481621"></div> <script src="https://patentbrief.org/embed.js" async></script>
Stay in the loop
Get a weekly digest of new patents.
One email per week. No spam. Unsubscribe anytime.
Keep exploring
Related patents you should know
US 4683195 · 1987
How to Make Billions of Copies of a DNA Segment
This patent describes the Polymerase Chain Reaction (PCR), a method to rapidly create many copies of a specific piece of DNA or RNA, enabling its detection and analysis.
Cetus Corp
US 8697359 · 2014
How to Edit Genes in Human Cells Using an Engineered CRISPR System
This patent describes an engineered CRISPR-Cas9 system for precisely cutting DNA in eukaryotic cells to change how genes work, opening the door for gene editing in complex organisms.
Massachusetts Institute of Technology
US 7657849 · 2010
How the iPhone's Slide-to-Unlock Gesture Works
Apple's 2010 patent describes unlocking a device by dragging a specific graphical image across the touchscreen along a predefined path, a gesture that became iconic with the original iPhone.
Apple Inc
US 4733665 · 1988
How Doctors Implant a Permanent Stent Using a Balloon
This patent describes the method for placing a permanent, expandable wire mesh tube inside a blood vessel or other body tube using a balloon-tipped catheter to widen it and keep it open.
Expandable Grafts Partnership
US 4965188 · 1990
How to Make Many Copies of a DNA Piece with Heat
This patent describes the Polymerase Chain Reaction (PCR) method, a technique to make millions of copies of a specific DNA segment using a heat-resistant enzyme and repeated temperature changes.
Cetus Corp
US 4235871 · 1980
How to Encapsulate Active Materials in Lipid Bubbles Efficiently
This patent describes a method for trapping biologically active substances inside tiny, multi-layered fat bubbles called liposomes, using a specific water-in-oil emulsion and gel-forming process to improve how much material gets captured.
Individual
Semantically similar
You might also find these interesting
US 9256215 · 2016 · Brain
How Artificial Neurons Learn from Spikes and Internal States
US 9256823 · 2016 · Qualcomm Technologies
Efficiently Updating Connections in AI Brains
US 20230409893 · University of Dayton
Training AI Chips Using Memristor Crossbars Directly On-Chip
US 8156057 · 2012 · Knowmtech
How Nanoparticles Form Adaptive Neural Network Connections
More to explore
More in AI & Machine Learning
US 10452978 · 2019 · Google LLC
How AI Models Understand Language Using 'Attention'
US 6523026 · 2003 · Huntsman International LLC
How Computers Find Hidden Connections Between Different Fields of Knowledge
US 11615208 · 2023 · Capital One Services LLC
How Cloud Systems Automatically Create and Train AI Data Models
US 10402750 · 2019 · Facebook Inc
How Facebook Uses Deep Learning to Predict What You Might Like
New to patents?
Common Questions
Frequently Asked Questions
What does How Brain-Inspired Computers Learn and Unlearn Information cover?
This patent describes a method for a brain-like computer network to learn associations between different inputs and then specifically unlearn mistakes or 'false positives' using a unique feedback mechanism.
Who owns patent US 11481621?
International Business Machines owns this patent, granted in 2022.
When does this patent expire?
This patent is expected to expire on August 16, 2039, when the invention enters the public domain.
What problem does this patent solve?
This technology is important because it addresses a fundamental challenge in artificial intelligence: making learning systems more efficient and robust. By providing a specific mechanism for 'unlearning,' it allows neural networks to refine their knowledge and correct mistakes, which is crucial for complex tasks. This approach, using brain-inspired 'spiking computation,' could lead to more energy-efficient AI hardware, especially for devices that need to learn on the go.
What does this patent NOT cover?
Does not cover traditional artificial neural networks that do not use 'spiking computation' or 'firing events' as described.
Same assignee
More from International Business Machines
Patent monitoring